Assume G is a simple graph with $n>3$ vertices. Prove that if the number of its edges is more than $\binom{n-1}{2} + 2$, then it has a Hamilton circuit.
Any hints?
Assume G is a simple graph with $n>3$ vertices. Prove that if the number of its edges is more than $\binom{n-1}{2} + 2$, then it has a Hamilton circuit.
Any hints?
On
First of all, notice that the number of the non-edges in the graph is at most $n-3$.
In the case of the complete graph the result is clear. Therefore suppose that the considered graph is not complete. Thus there are some pairs of vertices $u$ and $v$ such that $uv\notin E$. If for all such pairs $\deg(u)+\deg(v)\geq n$, then Ore's theorem solves the problem. Otherwise, there exists a pair of vertices, say $x$ and $y$ such that $xy\notin E$ and $\deg(x)+\deg(y)\leq n-1$.
Now, let us count the number of missing edges that are incident to $x$ or $y$. Clearly $xy$ is one such edge. From the remaining $2\cdot(n-2)$ possibilities at most $n-1$ is an edge, and at least $n-3$ is a non-edge in the graph. Thus, the number of missing edges is at least $1+(n-3)>n-3$, a contradiction. Thus, there are no such pairs $x,y$ for which $\deg(x)+\deg(y)<n$, therefore the graph does contain a Hamilton cycle.
Suppose that $\delta$ is the minimum degree of $G$. If $\delta =n-1$ or $\delta=2$, then there is nothing to prove. So suppose that $\delta \leq n-2$. Let $v$ be the vertex of degree $\delta$. In $G \setminus v$, the number of edges is $m'\geq \genfrac{(}{)}{0pt}{}{n-1}{2}-(\delta-2)$. We claim that $\genfrac{(}{)}{0pt}{}{n-1}{2}-(\delta-2) \geq \genfrac{(}{)}{0pt}{}{n-2}{2}+2$.
Because:
$\genfrac{(}{)}{0pt}{}{n-1}{2}-(\delta-2) \geq \genfrac{(}{)}{0pt}{}{n-2}{2}+2 \Leftrightarrow n-2 \geq \delta$.
So this is correct.
Now, by induction on $n$, $G\setminus v$ has a hamiltonian cycle $C$.
Suppose that the neighbors of $v$ are $v_1, ..., v_\delta$ and $v_i$ is adjacent $v_j$ by an edge $e$. If $e$ is on the cycle $C$, then we have a hamiltonian cycle in $G$. So suppose that non of edges between the neighbors of $v$ is on cycle $C$. Now, assume that the vertices before and after the vertex $v_i$ (clockwise), (in cycle $C$) is $v_{i_1}$ and $v_{i_2}$ and the similar name for $v_j$. If the vertices $v_{i_1}$ and $v_{j_1}$ or the vertices $v_{i_2}$ and $v_{j_2}$ are adjacent, then we can form a hamiltonian cycle. Therefore suppose that for any two pair of vertices of neighbors $v$, the graph $G\setminus v$ does not have these two kind of edges. So we have:
$m'\leq \genfrac{(}{)}{0pt}{}{n-1}{2} -2\genfrac{(}{)}{0pt}{}{\delta}{2}$.
So we have :
$2\genfrac{(}{)}{0pt}{}{\delta}{2} \leq \delta -2$.
And it is a contradiction.